Compass Investments

Etherscan has introduced the Build with AI toolkit, designed for AI agents that need to retrieve verified blockchain data in real time from 60 EVM-based networks.

The MCP server includes 20 tools for working with balances, transactions, smart contracts, and gas prices, while features such as Etherscan Flow and Contract Review help agents track funds and analyze contracts using only up-to-date API data.

This release aims to combat AI hallucinations related to wallet and transaction data; its launch coincided with the fact that traffic from bot agents recently surpassed human traffic on the network.

Etherscan has introduced a suite of tools that allows AI agents and development tools to retrieve real-time blockchain data directly from the explorers API.

The latest update is aimed at developers building agents that require verified blockchain data spanning more than 60 EVM-compatible networks. The company announced the launch of the Build with AI initiative in a post on X on August 31, 2026.

This update offers three options for machines to access Etherscan data. First, the Model Context Protocol (MCP) servera hosted interface that agents such as Claude and ChatGPT can query using natural language; second, a command-line tool for terminals, scripts, and CI pipelines; and third, a set of installable skills for agents.

According to Etherscans documentation, the MCP server operates on a single endpointmcp.etherscan.io/mcpand provides access to 20 tools.

These cover key blockchain read functions: balances in native currency and tokens, regular and internal transactions, transaction and receipt lookups, contract details, gas prices, and event logs. A single connection provides access to all supported networks.

The command-line interface (CLI) mirrors the API structure: commands follow the format `etherscan `, and output is provided in JSON format by default; in addition, there are options to output results as tables and CSV files to simplify working with the results.

Authorization for all these tools is handled via the standard Etherscan API key, and the free key works on all supported chains.

,detected_source_language:RU

The publicly available Etherscan repository on GitHub lists four skills that can be installed using the `npx skills add` command or by copying the folder to the agent’s skills directory.

The Etherscan Flow skill tracks the movement of funds between addresses and records this data in a case. The Etherscan Contract Review skill explains what a verified and deployed contract does. The transaction debugger reconstructs exactly what a given transaction did and why. The orchestrator skill routes the task to the appropriate interface.

According to the information in the repository, the main goal of these skills is to ensure that every address, amount, and transaction hash is obtained from an actual API call and is not fabricated. The entire set of skills is read-only and adheres to standard API request quotas and frequency limits.

The Etherscan release is seen as a solution to the problem of hallucinations, which has also crept into crypto AI tools.

Language models can fabricate a wallet balance or a transaction that never existed, and both can have serious consequences. This usually happens when they lack the necessary data, and in most cases, its because they cant index pages containing the information theyre looking for.

In its documentation, Etherscan also warned users not to enter their API keys into fake or fraudulent MCPs posing as legitimate ones. The company stated that there is only one official MCP server and warned that listings on MCP platforms calling themselves Etherscan MCP are in no way affiliated with it.

Thus, the scale of bot usage and activity requires appropriate measures to prevent oversights.

In early June, AI bots surpassed humans, becoming the largest source of traffic on the Internet. This was confirmed by Cloudflare CEO Matthew Prince, who stated that this happened sooner than he had predicted.

According to data from Cloudflare Radar, bots accounted for 57.4% of web traffic, compared to 42.6% from humans, with the share of bots even higher in North America68.6% .

At the moment, humans have regained the lead; however, over the past four weeks, bots have still accounted for more than 35% of web traffic.

“,”detected_source_language”:”RU